AI Adoption Update: Leaders Opening Up a Lead; Still Time to Catch Up
Early adopters that see value from AI systems invest more heavily in complex systems, opening up an innovation gap with companies not yet started with AI
By John P. Desmond, Editor, AI in Busines
AI adoption continues to accelerate at varying rates by organizations and industries, and with varying degrees of project success.
In a 2021 survey by PwC, answered by 1,036 high business titles from a range of industries, 86 percent of respondents reported that AI was becoming a “mainstream technology” at their company in 2021.
A number of responding companies had accelerated their AI investments in the midst of the COVID-19 surge. Now they are seeing benefits ranging from revenue growth to improved customer experiences. Companies who have rolled out AI enterprise-wide are more optimistic about growth, with 25 percent expecting to increase revenue, compared with 18 percent for all companies.
“The future payoff is even greater and could give early adopters an edge that competitors may never be able to overtake” the report authors state.
Similar results were reported from a study of how 950 companies in Australia are using AI, conducted by IDC and reported in Information Age. Some 39 percent of companies reported using AI. These users expected to increase their AI investments by an average of 34 percent annually through 2025, taking on use cases of larger scope and complexity in the process.
About 80 percent of the AI users reported that they applied AI to more than five use cases, and were seeing improvements in many areas, including employee productivity and customer experience. Retailers were using AI to optimize operations, improve customer service and improve decision-making around merchandise stock. Banks were leaning on AI to help with financial risk management, cybersecurity threat detection and automated compliance management.
“By 2024, 50 percent of our use cases and analytics will include AI to some degree,” stated Craig Turrell, head of P2P within the Digital Centre of Excellence at Standard Chartered Bank.
Despite its value, however, IDC found that fewer than 40 per cent of companies had an enterprise-wide strategy to coordinate their investments across the company.
Lourens Swanepoel, Australia data and AI lead with technology consultancy Avanade, suggested that executives need to actively improve what he calls the company’s AI-Q (AI knowledge and literacy) if they hope to tap AI’s benefits.
AI projects “often start off with big, grand intentions, but there is too much of a short-term, tactical view taken because it’s driven by just one use-case in the organization,” he stated, “and the challenge is how to scale it from there.”
Digital Factories Hot, But Only One in Five Have Implemented a Digital Twin
Digital factory initiatives incorporating AI are attracting investment, with $1.1 trillion annually being invested today, according to PwC’s 2022 Digital Factory Transformation Survey, as recently reported in RTInsights. The survey is based on responses from 700 global corporations from 23 countries.
Spoke 64 percent of companies report being in an early stage with digital transformation. For example, 19 percent of companies are operating with a fully-functional digital factory twin. And 48 percent report having fully implemented maintenance analytics. Compared to PwC’s first survey on the topic in 2016, factory digitalization is progressing much more slowly than anticipated, the RT authors stage.
Factoring holding back digital progress include: complex environments, highly diverse and distributed machine landscapes; and challenges associated with scaling across the entire production network.
“Despite slow progress in many cases, digitization is still high on the agenda,” stated Dr. Reinhard Geissbauer, author of the PwC study. He is a partner with PwC Management Consulting in Munich, with 18 years of experience in the industry. He added, “With new imperatives alongside cost and efficiency, due to the scale of external disruption in recent years, resilience and transparency are now the key drivers for digital transformation. Another driver that has been gaining importance, and will continue to do so, is sustainability.
Still Time to Catch Up
Maybe not being among the leading AI adopters today is not so dire; there is time to catch up. Companies considering how to implement AI can ask themselves several guiding questions, suggested Whit Andrews, an analyst with Gartner, in a recent account in InformationWeek,
First, identify who is going to sponsor the AI project and why it matters to the organization. For example, when the inevitable obstacles pop up–integration costs, staff availability, security concerns–it helps to have an executive sponsor. “C-suite executives have access to sources of funding and influence that may provide critical,” Andrews suggested.
They also know how to make a project relevant to the CEO. “Value is measured in impact to the aspects of the business that get attention,” he stated.
Next, will the project result in better skills, better data and better direction? Ideally, your AI projects have a theme that ties them together. Commitments are needed from data leaders, IT leaders and business leaders.
Lastly, consider whether AI is really needed for the project. This can be a tough decision to make; a bad bet can be costly. The average AI initiative that reaches production takes 7.3 months to get there, and 10 percent of initiatives take between one and two years, according to the 2021 Gartner AI in Organizations survey. And half of the initiatives take less than six months.
Andrews suggests that executives ask whether there is another way to do a project, without using AI. “If the answer is no, and if the project is strategic, then it’s time to get started,” he stated. If the project moves forward, he suggests setting up measurements for the cost of resources, identifying the challenge of getting it starting and accepted, and what ongoing effort is needed.
Lack of a Strong AI Culture “Generates Problems”
The Art of AI Maturity recent report from Accenture echoes similar results, with some nuanced differences. Accenture surveyed over 1,600 high-level executives and data science leaders from the world’s largest organizations. Some 75 percent were found to have integrated AI into their business strategies, and 12 percent are rated as using AI at a mature level, according to a recent account from Dice.
“The potential of the technology is still far from reaching its peak.” stated Ryohei Fujimaki, founder and CEO of dotData, author of the report. dotData offers an enterprise-grade AI automation platform.
A major reason organizations struggle to move AI projects into production, is the lack of a strong AI culture. “This generates all kinds of problems,” Fujimaki stated.
Any CEO would want to exploit the benefit of an AI system, such as models that optimize marketing spending by tuning the mix, or models that enhance revenue by lowering customer churn. Whether they could be implemented, depends on having sufficient data, which may not be available. That’s a limitation of AI.
It’s also important for business managers to understand how much time it takes to build a machine learning model; it takes between 31 and 90 days for most companies, according to the “2020 State of Enterprise ML” report from Algorithmia. Of 745 companies surveyed, 63 percent had deployed machine learning models into production. Some 40 percent of the companies reported it takes more than a month to deploy an ML model into production.
On the technical side, things can go wrong when the pilot, based on a subset of data, tries to scale to enterprise-wide data. “The model may drift, fail or collapse,” Fuimaki stated. Also, data scientists may code in Python, which is a high-level, interpreted, general-purpose programming language. It may have to be converted to a language that will work better in the production system, a time-consuming and difficult task.
Read the source articles and information from PwC, in Information Age, in PwC’s 2022 Digital Factory Transformation Survey, in RTInsights, in InformationWeek, in the Art of AI Maturity report from Accenture, in Dice and in the 2020 State of Enterprise ML report from Algorithmia.
(Write to the editor here.)